Re: Development question regarding new resources
From: Robert Stankey (robert.stankey_at_lsil.com)
Date: 03/18/04
- Next message: Steve T.: "RE: Cannot "Manage Server" because system thinks it's a Cluster."
- Previous message: Greg Page: "RE: 3 Node Cluster With SQL 2000 AND Exchange 2003"
- In reply to: Greg Page: "RE: Development question regarding new resources"
- Messages sorted by: [ date ] [ thread ]
Date: Wed, 17 Mar 2004 22:47:53 -0600
On Thu, 18 Mar 2004 03:09:41 GMT, Greg Page wrote:
I'm trying to get a node to failover if it loses access to the storage.
When this happens I'll see a SelectionTimeout error returned but
clusdisk.sys doesn't seem to recognize this - or more specifically it
doesn't appear to initiate a node failover.
> What exact.y are you trying to accomplish? what is your end goal?
>
> This message is provided as is and does not imply any warranty or gaurantee
> of support of any kind. Microsoft, the person providing the advice, nor any
> of Microsoft's and posters subsidiaries, friends, pets, children or
> acquaintenances will be held in any way liable for following any advice
> contained herein.
> --------------------
>>From: Robert Stankey <robert.stankey@lsil.com>
>>Subject: Development question regarding new resources
>>User-Agent: 40tude_Dialog/2.0.10.1
>>MIME-Version: 1.0
>>Content-Type: text/plain; charset="us-ascii"
>>Content-Transfer-Encoding: 7bit
>>Organization: LSI Logic
>>Date: Wed, 17 Mar 2004 11:19:58 -0600
>>Message-ID: <ffo59vbd9yht$.iwtbwhaobszi$.dlg@40tude.net>
>>Posted-And-Mailed: yes
>>Newsgroups: microsoft.public.windows.server.clustering
>>NNTP-Posting-Host: proxy1.lsil.com 147.145.40.41
>>Lines: 1
>>Path: cpmsftngxa06.phx.gbl!TK2MSFTNGP08.phx.gbl!TK2MSFTNGP11.phx.gbl
>>Xref: cpmsftngxa06.phx.gbl microsoft.public.windows.server.clustering:2973
>>X-Tomcat-NG: microsoft.public.windows.server.clustering
>>
>>I've seen problems with the ScsiPort driver not properly recognizing Fibre
>>Channel RAID controllers when they log themselves into the fibre. To work
>>around this some HBAs allow you to disable the bus rescan notification from
>>the miniport driver. One problem with doing this is that if you disconnect
>>the fibre cables from the host Cluster won't failover to the other node
>>because I believe its relying on the Pnp REMOVE_DEVICE notification to be
>>sent down the driver stack.
>>
>>Because of this dilema I consulted the Server Cluster API documentation and
>>noticed you can write new Resource DLL's so in theory I could write a
>>resource to monitor the disks but that seems redundant. As part of this
>>solution I have a filter driver that can determine if the connection to the
>>storage is lost by examining any I/O errors returned. Given this my new
>>DLL could poll the filter driver during the DLL's LooksAlive() or IsAlive()
>>routine and return status/failure.
>>
>>So, my questions are:
>>
>>- Is there anyway to "piggyback" on the Disk resource so a new, customized
>>version is not required?
>>- I assume an adminstrator would have to associate an instance of my
>>Resourcce with each Disk, or could one instance be created which could
>>monitor multiple devices?
>>
>>Thanks,
>>Bob
>>
- Next message: Steve T.: "RE: Cannot "Manage Server" because system thinks it's a Cluster."
- Previous message: Greg Page: "RE: 3 Node Cluster With SQL 2000 AND Exchange 2003"
- In reply to: Greg Page: "RE: Development question regarding new resources"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
